When we perform a motor task, unexpected perturbations produce errors in the intended sensory consequences of our movements. Behavioral studies have suggested that the nervous system updates its motor commands in order to minimize these `sensory predictions errors' on a trial-to-trial basis, a phenomenon known generally as motor learning. By abstracting psychophysical results into scalar variables, state-space models have proven useful in explaining the relationship between error and learning. However, a fundamental limitation of these models in the interpretation of motor behavior, is that movements involve complex activation of muscles, patterns that are outside the scope of current scalar models that have been used to represent learning and error. In order to investigate the mechanism of trial-to-trial motor learning we must move beyond the state-space framework and consider how these temporal patterns of muscle activation are altered after experiencing a perturbation. Here, we propose that the motor commands that the nervous system produces in response to perturbations, i.e., the feedback responses that correct for the prediction error during a movement, also serve as a teacher for the motor learning process. The goal of our proposal is to use electromyography (EMG) to systematically evaluate the relationship between feedback responses and motor learning. We will test our proposed hypothesis in three ways. First, we will systematically alter the shape of the feedback response (through manipulation of perturbations), and test whether this alteration results in consistent modulation of trial-to-trial motor learning. Second, we will modulate the gain of the feedback response (through manipulation of task parameters), and test whether this gain modulation results in corresponding changes in trial-by-trial motor learning. Third, we will increase or decrease the amount that the brain learns from error (through manipulation of error history), and test whether this modulation of learning coincides with modulation of the feedback response. Finally, we will stimulate the cerebellum, a region that is thought to be critical for motor learnig, and determine to what extent this neural substrate affects the relationship between feedback responses and trial-to-trial learning. From a clinical standpoint, understanding the relationship between feedback and learning has direct implications in the design of motor rehabilitation paradigms for neurotrauma or disease. The results of this work may suggest that rehabilitation strategies should allow the nervous system to make predictions and correct for its errors, in order to more effectively promote motor learning. In addition, the learning-feedback correlation and its relation to error-sensitivity may offer insight into potential rehabilitation techniques tht could enhance the rate of learning outside the clinic. Finally, our investigation of cerebellar involvement in learning and feedback may improve our understanding of cerebellar ataxias. We may also identify methods to alter the learning-feedback response relationship using neural stimulation, which could enhance therapeutic benefits of rehabilitation paradigms.